Drosophila nonmuscle myosin II is required for rapid cytoplasmic ...

Development 121, 1937-1946 (1995) Printed in Great Britain © The Company of Biologists Limited 1995

1937

Drosophila nonmuscle myosin II is required for rapid cytoplasmic transport during oogenesis and for axial nuclear migration in early embryos Sally Wheatley1,*, Sanjay Kulkarni2 and Roger Karess1,† 1C.N.R.S. Centre de Génétique Moleculaire, Avenue de 2Department of Biochemistry, NYU Medical Center, 550

la Terrasse, 91198 Gif-sur-Yvette, France First Avenue, New York, NY 10016, USA

*Present address: Worcester Foundation for Experimental Biology, Shrewsbury, MA 01545, USA †Author for correspondence

SUMMARY The X-linked Drosophila gene spaghetti squash (sqh) encodes the regulatory light chain of nonmuscle myosin II. To assess the requirement for myosin II in oogenesis and early embryogenesis, we induced homozygous germline clones of the hypomorphic mutation sqh1 in otherwise heterozygous mothers. Developing oocytes in such sqh1 germline clones often failed to attain full size due to a defect in ‘dumping’, the rapid phase of cytoplasmic transport from nurse cells. In contrast to other dumpless mutants described to date, sqh1 egg chambers showed no evidence of ring canal obstruction, and no obvious alteration in the actin network. However the distribution of myosin II was abnormal. We conclude that the molecular motor responsible for cytoplasmic dumping is supplied largely, if not

exclusively, by nurse cell myosin II and we suggest that regulation of myosin activity is one means by which cytoplasmic transport may be controlled during oocyte development. The eggs resulting from sqh1 clones, though smaller than normal, began development but exhibited an early defect in axial migration of cleavage nuclei towards the posterior pole of the embryo, in a similar manner to that seen in early cleavage eggs in which the actin cytoskeleton is disrupted. Thus both nurse cell dumping and axial migration require a maternally supplied myosin II.

INTRODUCTION

zipper gene (zip), encoding the MHC, has demonstrated the importance of myosin II in the changes of cell shape accompanying the tissue movements of dorsal closure in the late embryo (Young et al., 1993). A severe, though hypomorphic, allele of the spaghetti squash gene (sqh) encoding the RMLC, permitted survival through the larval period of the fly life cycle, but caused frequent failure of cytokinesis in the dividing cells of larval imaginal tissues (Karess et al., 1991). These genetic studies of zygotic mutants do not exclude the participation of mysoin II in other aspects of Drosophila development, however, such as those events occurring after the lethal phase of the mutations. Moreover, the effect of myosin II mutations on events prior to the lethal phase may be masked by the maternal contribution of wild-type myosin II to the egg cytoplasm. Indeed, myosin II exhibits pronounced tissuespecific changes in its subcellular location throughout fly embryogenesis (Young et al., 1991), which often correlate with periods of cellular and subcellular movements. In the present study we genetically assess the need for myosin II in two other developmentally important aspects of cytoplasmic motility in Drosophila: the movement of nuclei during the syncytial stages of the early embryo, and the rapid intercellular transport of cytoplasm leading to nurse cell regression observed during oogenesis. Immediately following fertilization, the rapidly cleaving

Changes in the microfilament and microtubule cytoskeleton underlie essentially all of the movements of developing tissues. Dissecting the roles of these filaments and their accompanying molecular motors in cytoplasmic motility is therefore crucial to understanding the forces involved in morphogenesis. The actin-dependent molecular motors, the myosins, are likely to participate in many aspects of cytoplasmic movement (reviewed by Cheney and Mooseker, 1992; Titus, 1993), but the specific functions carried out by individual myosin species during the cellular movements of development are for the most part still to be determined. In the fly Drosophila melanogaster the isolation of mutations in two of the three genes encoding polypeptides which make up nonmuscle myosin II has made it possible to examine genetically which particular events in the development of a metazoan require this specific myosin motor. Myosin II is a hexamer consisting of a pair of heavy chains (MHCs) carrying the motor domain and the tail, and pairs of the essential and regulatory light chains (EMLCs and RMLCs, respectively) (reviewed by Korn and Hammer, 1988). The RMLC is a critical component of myosin II which normally regulates myosin activity according to its state of phosphorylation (reviewed by Sellers, 1991). A null mutation in the

Key words: myosin regulatory light chain, spaghetti squash, nurse cell dumping, cell motility, Drosophila, oogenesis

1938 S. Wheatley, S. Kulkarni and R. Karess nuclei of the early Drosophila embryo are initially positioned anteriorly in the egg. During mitotic cycles 4 through 7, the nuclei migrate as if they were on the surface of a narrow ellipsoid expanding axially towards the posterior end, while remaining deep beneath the surface (Zalokar and Erk, 1976; Baker et al., 1993). Subsequently, during cycles 8 and 9, the nuclei migrate outwardly towards the cortex where they eventually uniformly populate the subcortical surface of the egg (Foe and Alberts, 1983; Baker et al., 1993). The first nuclear movement, termed axial expansion, is known to be dependent on a functional actin cytoskeleton, as it is sensitive to treatment with actin poisons such as cytochalasin-D, while the subsequent cortical migration requires microtubules and is blocked by colchicine (Zalokar and Erk, 1976; Karr and Alberts, 1986; Hatanaka and Okada, 1991; reviewed by Foe et al., 1993) Nearly all the cytoplasmic components of the early embryo are supplied maternally during oogenesis. In the developing egg chamber, 15 interconnected nurse cells transfer their cytoplasm to the oocyte through cytoplasmic bridges called ring canals. This process occurs slowly throughout oogenesis up to stage 10 (see Spradling, 1993, for a description of stages and review of oogenesis), but during stage 11, the entire cytoplasmic content of the nurse cells is transferred to the oocyte within 30 minutes, in a step known as ‘dumping’. As shown by drug studies (Gutzeit, 1986), as well as by certain mutations in genes encoding actin-associated proteins (Cooley et al., 1992; Xue and Cooley, 1993; Cant et al., 1994; Verheyen and Cooley, 1994), dumping depends on an intact actin cytoskeleton. Both these events, axial expansion of cleavage stage nuclei and cytoplasmic dumping during oogenesis, are developmentally important examples of actin-based motility, for which the specific myosin motor, if any, has not been identified. In the present study, we have taken advantage of the hypomorphic sqh1 mutation to generate homozygous sqh1 mutant germline clones in otherwise heterozygous females. In this manner we were able to assess the requirement for functional myosin II during oogenesis and early embryogenesis of Drosophila. We find that while much of oogenesis proceeds normally, the rapid phase of cytoplasmic transport characteristic of stage 11 egg chambers is retarded and inefficient in the sqh1 mutant clones. Later, in clonally derived fertilized eggs, we find that the axial expansion of the cloud of cleavage nuclei during the first 6 cycles of nuclear division is also perturbed. Thus, both dumping and axial expansion require functional myosin II supplied by the maternal germ line.

Generation of germ-line clones The FLP-DFS system of Chou and Perrimon (1992) was used to generate homozygous sqh1 germline clones. Females y w sqh1 sn3 FRT101 /FM7 were crossed to ovoD1 FRT101/Y; hs-flp-F38 males. 24-hour pulses of eggs from this cross were allowed to develop to 2nd or early 3rd instar larvae and then heat shocked at 37°C for 2 hours. For some studies younger or older individuals were heat shocked. Heat shock-induced expression of the FLP recombinase leads to mitotic recombination between the two FRT sites on the X chromosome homologs. A fraction of the cells thus become homozgyous sqh1 ovo+ while all others remain ovoD1. The ovoD1 mutation causes dominant female sterility by blocking oogenesis at a very early stage. Non-FM7 adult females, nominally ovo+ sqh1 sn3 FRT/ovoD1sqh+ sn+ FRT, were collected and crossed to wild-type males. The successful induction of mitotic recombination in the somatic cells of these flies could be easily monitored by the frequent appearance of y sn bristles on the thorax and abdomen. These females were allowed to lay eggs or were dissected for examination of their ovaries. Any developing egg chambers had lost the ovoD1 allele and therefore were genetically sqh1/sqh1. The eggs produced from these germline clones are called ‘sqh1 eggs’ (without italics), to indicate the phenotype of the egg cytoplasm. However, their zygotic genotype depends on the fertilizing sperm. They may be either sqh1/Y males or sqh1/+ females.

MATERIALS AND METHODS

Immunocytochemistry Ovaries were dissected in phosphate-buffered saline (PBS, pH 7.2) to release the egg chambers, and transferred within 30 minutes to 10% formaldehyde in PBS. After fixing in formaldehyde for 15 minutes they were washed in PBS and left overnight at 4°C in PBS with 1% Triton X-100. Staining was achieved as detailed for embryos. Eggs were collected at 25°C on agar-sugar plates supplemented with yeast paste at the intervals stated. After the appropriate incubation period embryos were dechorionated in bleach (5 minutes at room temperature), washed throroughly in distilled water, and treated for 1 minute with heptane. The solution was replaced with a bilayer of heptane and 8% formaldehyde in PBS and incubated for 0.5-1.5 hours at at room temperature. The vitelline membrane was subsequently removed manually in the fixative using tungsten needles. Embryos were then washed in PBS and incubated overnight at 4°C in PBS with 1% Triton X-100. (When studying microtubules the PBS was supplemented with 50 mM EGTA and 5 µM taxol.) All antibody incubations were performed at 37°C for 1 hour in PBS, 0.3% Triton X-100, 3% normal goat serum. Anti-cytoplasmic myosin heavy chain, a gift from Dan Kiehart (Kiehart et al., 1990) was used at 1/100. Secondary antibodies (obtained from Jackson Immunoresearch) were routinely used at 1/50 or 1/100 dilution. Rhodamine or fluorescein-conjugated phalloidin (Molecular Probes) was used at 0.2 µM. Nuclei were stained with DAPI at 0.5 µg/ml. We used an antidetyrosinated α-tubulin monoclonal antibody, ID5 (Wehland and Weber, 1987), to label the sperm tail within the embryo (used at 1/20 dilution). This served both to identify the fertilized eggs and also to mark unequivocally the anterior end of the embryo which otherwise was difficult to determine due to the abnormal morphology of sqh1 eggs.

Fly stocks Flies were raised on standard corn meal Drosophila medium at 2526°C. The markers and balancers are described by Lindsley and Zimm (1992), except as noted. The isolation and cloning of sqh1 is described byKaress et al. (1991). In the text, the terms sqh gene product and RMLC are used interchangeably. The stocks of FRT101 and ovoD1 FRT101/Y; hs-flp-F38 used to generate germline clones (Chou and Perrimon, 1992) were obtained from Dr N. Perrimon. The sqh1 allele was crossed on to the FRT101-containing chromosome by standard recombination. Wild-type controls used the stock y w67c, obtained from the Bloomington Stock Center.

Microscopy All samples were viewed using a Leitz DMRB (Leica) microscope fitted with rhodamine, fluorescein and UV channels for fluorescence microscopy and capable of phase and Nomarski images. Samples were routinely mounted in 90% glycerol with 1% phenylenediamine (antifade) in pitted slides and viewed using dry lenses at 10× and 20×, and under oil at 40× or 63× magnification. For observations on live embryos, undechorionated eggs were rendered translucent by submerging them in halocarbon oil. The overlying coverslip was supported by two coverslip fragments to prevent it from touching the eggs.

Myosin II in oogenesis and early cleavage 1939 Definition of late egg chamber stages The morphological markers allowing a wild-type egg chamber to be staged are not always consistent in mutants of oogenesis. Thus in some sqh1 egg chambers, features of stage 10 or 11 (based on the ratio of nurse cell cytoplasm to oocyte cytoplasm) can be found coexisting with landmarks of stage 13. In staging these mutant egg chambers, we relied principally on the landmarks of follicle cell activity. Stage 10A: follicle cells around oocyte are still columnar. Stage 10B: transverse actin filaments within the nurse cells become apparent with fluorescein-phalloidin staining. Stage 11: follicle cells have begun to flatten. Stage 12: the region of dorsal appendage assembly can be seen by fluorescein-phalloidin staining of actin, but the appendages themselves are not visible by light microscopy. Stage 13: the dorsal appendages are visible by light microscopy, but are incomplete. Stage 14: the eggshell is essentially complete. See Spradling (1993) for a detailed description of the stages of oogenesis.

RESULTS Maternally supplied sqh+ product is required for normal development sqh1 is a hypomorphic allele of the gene encoding the RMLC of nonmuscle myosin II, reducing by greater than 90% the amount of message (Karess et al., 1991) and protein (S. B. Kulkarni, and R. Karess, unpublished), but not altering its structure. Employing the method of Chou and Perrimon (1992), which combines the high frequency inducible mitotic recombination of the yeast FRT-FLP recombinase system with the dominant female sterile technique for identifying germline clones (Perrimon and Gans, 1983), we generated sqh1/sqh1 mutant germline clones in approximately 90% of the otherwise sqh+ ovoD1/sqh1 ovo+ heterozygote females. This allowed us to monitor the requirement for RMLC, and thus myosin II, in oocyte development and early embryogenesis. The germline clones did indeed produce eggs (hereafter referred to as ‘sqh1’ eggs) which tended to be smaller and more fragile than those of wild type. The actual egg length varied considerably: the smallest ones were reminiscent of those found in such mutants of Drosophila oogenesis as singed, chickadee, and kelch (see Spradling, 1993), while others were nearly wild type in size. Eggs derived from sqh1 mutant female germlines invariably failed to hatch. Examination of fixed and stained eggs showed that 66% (n=193) of sqh1 eggs had begun development, compared to 96% (n=82) of wild-type eggs. Note that the sqh1 germline-derived embryos, fertilized by wild-type sperm, are genotypically either sqh1/Y males or sqh1/+ females. Normal embryos of these genotypes (that is, derived from sqh1/+ maternal germ cells) would have hatched and grown at least to the third instar larval stage, since zygotic sqh1 mutant animals survive up to the late larval period and manifest the mutant sqh1 phenotype (polyploidy in imaginal tissues) only at the larval stage (Karess et al., 1991). The fact that, regardless of their genotype, these embryos do not survive demonstrates an essential requirement for maternally derived sqh+ gene product, and therefore myosin II, for development during embryogenesis. Morphology of sqh1 egg chambers indicates a failure of nurse cell cytoplasmic transport As perturbation of egg quality signifies problems during

oogenesis, we examined developing sqh1 egg chambers by light and fluorescence microscopy. Prior to stages 10B-11 there was no consistent feature that could distinguish the majority of sqh1 egg chambers from wild type. The first consistent gross defects of sqh1 egg chambers were seen during stages 11-12, the period of dumping, when the oocyte normally expands at the expense of the regressing nurse cells. In wild-type egg chambers this is a rapid synchronous event whereby all the nurse cell cytoplasm is transferred within 30 minutes into the developing oocyte, resulting by the end of stage 12 in a compact apex of nurse cell nuclei at the anterior of the oocyte. As a consequence of the speed at which this event occurs, wildtype egg chambers in stage 11 are relatively rare. It was immediately apparent that many sqh1 egg chambers seemed to be blocked in stage 11 (Fig. 1). If stage 11 is defined simply as when the nurse cells still retain some cytoplasm, but their combined volume is smaller than that of the oocyte, then among wild-type chambers in developmental stages 10-12, 67% were at stage 10, 9% at stage 11, and 23% at stage 12. By contrast, 59% of sqh1 egg chambers were at stage 10, 38% at stage 11 and only 3% at stage 12. Closer examination of the sqh1 egg chambers however revealed many to have begun the assembly of the dorsal appendages, chorionic structures normally first seen in stage 13 egg chambers (Fig. 1A,B,D), indicating that these egg chambers are considerably more advanced than stage 11. Indeed, some egg chambers classified as stage 10 according to the relative volume of oocyte and nurse cells were also found with growing dorsal appendages (Fig. 1D), and thus were clearly much older. Fig. 2 compares wild-type and mutant egg chambers, both just initiating dorsal appendage growth (detectable at this time only by the outline of the surrounding follicle cells). The wild-type chamber has neatly compacted nurse cell nuclei clustered at the anterior apex, while in the sqh1 egg chamber, the nurse cells have retained much of their cytoplasm. Thus the transport of cytoplasm to the oocyte in sqh1 egg chambers is often incomplete, precisely the phenotype described for the ‘dumpless’ class of oogenesis mutants (Spradling, 1993; Cooley et al., 1992). Actin structures are normal, but myosin II distribution is abnormal in sqh1 egg chambers Prior to stage 10B, the major F-actin-containing structures in the nurse cells of wild-type egg chambers are the intercellular ring canals and the subcortical actin associated with the plasma membranes (Warn et al., 1985; Gutzeit, 1986). Starting at stage 10B, thick transverse microfilament bundles form a ‘halo’ around the nurse cell nuclei and extend towards the plasma membrane (Gutzeit, 1986), apparently serving to anchor the nuclei (Cooley et al., 1992). When visualized with fluorescein phalloidin, these three major F-actin populations are all essentially normal in sqh1 egg chambers (compare Fig. 3B and E; also visible in Fig. 2D). Myosin II was localized in egg chambers using a well-characterized polyclonal anti-myosin heavy chain (MHC) antibody (Kiehart and Feghali, 1986; Young et al., 1991). At stage 10, the subcortical cytoplasm of wild-type nurse cells showed strong anti-MHC staining (Fig. 3C), coinciding with the subcortical actin. The follicle cells, which completely envelope the developing oocyte and nurse cells, are also strongly staining along their borders and in the perinuclear region (Fig. 3C).

1940 S. Wheatley, S. Kulkarni and R. Karess

Fig. 1. sqh1 egg chambers show a dumpless phenotype. Normarsky images showing retention of nurse cell cytoplasm at late stages of sqh1 egg chamber development. All egg chambers are oriented with anterior to the right except the lower chamber in C. (A) A stage 13 egg chamber, note dorsal appendage and micropyle (small arrowhead). (B left) A stage 13 egg chamber. (B right) A string of early egg chambers with normal morphology. (C top) A stage 10 egg chamber of normal morphology. (C bottom) an essentially complete oocyte (stage 14) with short stubby dorsal appendages. (D) A stage 13 egg chamber, with a ratio of nurse cell to oocyte cytoplasm characteristic of stage 10. Arrowheads in A,B,D indicate dorsal appendages. Scale bar, 100 µm.

Fig. 2. Comparison of wild-type and sqh1 egg chambers at the beginning of dorsal appendage development (late stage 12 or early stage 13). (A,B) Wild type. (C,D) sqh1. (A,C) DAPI staining of DNA. (B,D) Fluorescein-phalloidin staining of F-actin filaments. In all figures anterior is to the right. Scale bar, 50 µm.

Myosin II in oogenesis and early cleavage 1941 Myosin was not found associated with the transverse actin filaments of the nuclear halo nor with the ring canals. The most notable feature labeled by anti-MHC in sqh1 mutant egg chambers was a population of strongly staining particles, mostly at the surface but also deeper within the nurse cells (Fig. 3F,G), containing little or no actin. Because this staining was restricted to sqh1 mutant tissue, and since similar particles have also been observed in somatic tissues of homzygous sqh1 larvae (data not shown), we believe it represents nonfunctional aggregates of myosin II, resulting from the dearth of regulatory light chain (Trybus et al., 1994). Otherwise, the anti-MHC labelled the same structures in the

subcortical region of the sqh1 nurse cells as in wild-type egg chambers (compare Fig. 3F,G with 3C). The follicle cells, which are of somatic origin (and therefore wild type), also retain the wild-type pattern of anti-MHC staining. Abnormal nurse cell number in sqh1 egg chambers A significant fraction of sqh1 egg chambers had an abnormal number of nurse cell nuclei (Fig. 4 and Table 1). 21% had 13 nuclei or fewer, some as few as 9, while 6% of the egg chambers had approximately 30, double the normal number, but such chambers always included a developing oocyte at the posterior end (Fig. 4C,D). In addition, binucleate nurse cells

Fig. 3. sqh1 stage 10B egg chambers have normal actin structures, but myosin II distribution is abnormal. Anterior is up in all frames. (A-C) Wild-type egg chambers. (D-G) A sqh1 egg chamber. (A) A wild-type egg chamber labeled with the DNA stain DAPI. (B) A similar egg chamber stained with fluoresceinphalloidin to label F-actin. (C) Anti-MHC immunostaining of the same egg chamber as in A. Myosin is present predominantly in the subcortical region of the nurse cells and in the follicle cells. The very fine whispy anti-MHC staining (small arrowheads in C and F) marks the outline of the follicle cells which envelope the anterior egg chamber. The faint rings of myosin in C are not ring canals, but rather correspond to the perinuclear region of the follicle cells. (D,E) DAPI and fluorescein-phalloidin staining, respectively, of a single stage 10B sqh1 egg chamber. In the nurse cells of sqh1 egg chambers the subcortical actin (c), the transverse actin filament bundles (f), and the ring canals (r), all appear normal. (F,G) The same egg chamber showing surface and more medial focal planes, respectively, of anti-MHC staining. Numerous MHC-containing particles are found exclusively in the sqh1 nurse cells. Scale bar, 25 µm

1942 S. Wheatley, S. Kulkarni and R. Karess were occasionally seen (Fig. 4A,B,E,F). The frequency of binucleate cells was not measured; the fraction of abnormal egg chambers is probably therefore an underestimate.

originally juxtaposed to the vitelline membrane, contracts to create a clear zone into which the pole cells subsequently bud. This cortical contraction is an actin-dependent process, and is closely associated with the process of axial expansion (see Foe et al., 1993, for review). We examined posterior zone clearing in newly laid wild-type and sqh1 eggs, rendered translucent in halocarbon oil, over a 2-hour period. Under these conditions 75.6% (n=45) of wild-type eggs exhibited a visible posteriorzone clearing. Given the fraction of of sqh1 eggs starting to develop (66%) and the fraction of wild-type eggs showing posterior clearing, one would expect approximately 50% of sqh1 clones to show posterior clearing. In fact, only 7.9% (n=38) of sqh1 eggs did so, indicating that this process is indeed defective.

Nuclear migration is perturbed in sqh1 embryos As detailed in the Introduction, the first 3 mitotic divisions following fertilization occur in the anterior third of the egg. During the next 4 division cycles, the cloud of nuclei expands towards the posterior in a process referred to as ‘axial expansion’ (Zalokar and Erk, 1976; Baker et al., 1993). Subsequently, the nuclei migrate outwards towards the cortex during mitoses 8 and 9, and uniformly populate the blastoderm surface by cycle 10 (Foe and Alberts, 1983). In sqh1 eggs, the early cleavage nuclei retained an anterior bias within the egg and engaged prematurely in cortical migration to the periphery (Fig. 5). Nuclear migration was often more isotropic, producing a spherical cloud of nuclei, DISCUSSION with the result that within the ellipsoid of the egg, many nuclei quickly reached the subcortical surface, while the posterior end Myosin II, the classical long-tailed cytoplasmic (nonmuscle) of the embryo remained devoid of nuclei (Fig. 5C,D). That myosin, has long been suspected of having a role in many cortical migration was occuring prematurely can be seen by motile events in eukaryotic cells. However, the realization that comparing the wild-type egg in Fig. 5A with the sqh1 egg of Fig. 5B. Although both are in cycle 7, the nuclei of the sqh1 embryo have nearly reached the surface, while the nuclei in the wild-type embryo remain deep within the yolk. In addition to the anterior bias, the placement of nuclei along the subcortical surface was often irregular (Fig. 5E). Once at the surface, compensating divisions at the posterior front of nuclei tended to fill most of the posterior void, (Fig. 5E,F) and at later time points (3-3.5 hours after egg laying), the distribution of nuclei in some embryos was near wild type. Staining with the anti-centrosome antibody RB188 (Whitfield et al., 1988) revealed that most of the nuclei retained their centrosomes, although occasional free centrosomes were also seen in local bald patches in the nuclear pattern (data not shown). Some brightly staining material was detected in the posterior void, but did not have the form of centrosomes. There was no suggestion that the centrosomes could move posteriorly independently of the nuclei. Many of these embryos succeeded in at least partially cellularizing, particularly in the anterior half of the egg. However cellularization in sqh1 eggs was not examined in detail in this study as the preceeding perturbations of nuclear distribution rendered difficult the interpretation of the abnormal cellularization. Fig. 4. Nurse cell number can vary in sqh1 egg chambers. All panels are sqh1 egg chambers. Posterior zone clearing often fails in sqh1 eggs During early cleavage stages in wild-type embryos, the yolky posterior cytoplasm,

(A,C,E) DAPI staining of DNA. (B,D,F) Fluorescein-phalloidin staining to outline nurse cells. (A,B) A stage 10 egg chamber with at least one binucleate cell (E,F) A stage10 egg chamber with only 10 nurse cell nuclei distributed in 8 nurse cells. (C,D) A stage 10 egg chamber with approximately thirty nuclei. Arrowheads indicate pairs of nuclei sharing single nurse cells. Scale bar, 100 µm.

Myosin II in oogenesis and early cleavage 1943 Table 1. Variable numbers of nurse cell nuclei are found in sqh1 egg chambers Nurse cell nuclei per egg chamber

Wild type (n=62) sqh1 (n=70)

14/15*

13

12

11

10

9

>15

62 51 (73%)

0 5

0 3

0 1

0 5

0 1

0 4 (6%)

*Since even in wild-type chambers it was sometimes difficult to reliably identify all 15 nurse cells , abnormal egg chambers were conservatively defined as those having 13 or fewer nuclei.

myosin II is just one member of a family of actin-dependent molecular motors has necessitated a proper assignment of the specific role each myosin plays in any given aspect of actinbased cytoplasmic motility. In this report we have shown that two specific actin-dependent subcellular movements important for Drosophila development, the rapid cytoplasmic transport from the nurse cells to the developing oocyte, and the axial migration of nuclei in the early embryo, require a functional myosin II. Myosin II in oogenesis The fast transport or ‘dumping’ of the nurse cell cytoplasm is inefficient in sqh1 egg chambers. This was manifest by the large number of egg chambers in which relatively full nurse cells coexisted with developing dorsal appendages, a hallmark of stage 13. Several other mutations have been reported to produce defects in cytoplasmic dumping during oogenesis. In females homozygous for certain alleles of the genes chickadee (Cooley et al., 1992), which encodes profilin, or singed (Cant et al., 1994) which encodes a homolog of the actin-bundling protein fascin, or quail (Mahajan-Miklos and Cooley, 1994), encoding a villin-like molecule, there is no assembly of the stage 10B actin filament bundles radiating from the nurse cell nuclei. Without these filaments, the nurse cell nuclei are untethered and soon plug the ring canals, blocking subsequent cytoplasmic transport. The mutations of the kelch gene (Xue and Cooley, 1993) disrupt maintenance of the ring canals themselves, again resulting in a physical block to transport. Germline clones of mutations in the armadillo gene, encoding a plakaglobinrelated molecule, cause severe defects to all actin structures, the cortical actin, the stage 10B filament bundles, and even the ring canals (Peifer et al., 1993), and dumping is often blocked in such chambers. In contrast to these mutants, the sqh1 germline clones do not alter the appearance of the actin structures in any profound manner. The overall morphology of the egg chambers up to stage 10B and the disposition of both the subcortical actin and the perinuclear halo of actin filaments are apparently normal. The myosin II distribu-

tion, however, is abnormal. Anti-MHC staining revealed particulate structures near the surface of sqh1 egg chambers. We suggest that these particles are clumps of nonfunctional myosin II lacking regulatory light chain, as biochemical studies have demonstrated a pronounced tendency for myosins stripped of their light chains to aggregate (Trybus and Lowey, 1988; Trybus et al., 1994). It is likely that all the phenotypes associated with sqh1 stem from an effective reduction in the amount of functional myosin II in the cell due to these aggregates. Thus

Fig. 5. Axial expansion fails to occur in sqh 1 clonally derived embryos. Anterior is to the left in all images. (A) Wild-type cycle 7 embryo. (B-F) sqh 1 clonally derived embryos. (B) cycle 7 embryo. The nuclei are clustered in the anterior half, and have already approached the cortical surface, while in A the nuclei remain in the interior and have spread the length of the egg. (C,D) Surface and medial views, respectively, of a sqh 1 egg showing the spherical distribution of nuclei. (E) Slightly older sqh 1 egg showing an irregular distribution of nuclei in the anterior half, with a few nuclei in the posterior region. (F) A sqh1 embryo with a near normal anterior syncytial blastoderm appearance. Scale bar, 100 µm

1944 S. Wheatley, S. Kulkarni and R. Karess the failure of sqh1 egg chambers to efficiently dump their cytoplasm does not appear to be a secondary consequence of blocking the cytoplasm ‘pipeline’, but is more likely to be a direct result of an insufficient motor activity normally provided by myosin II. The participation of subcortical actin-myosin contractions in the fast transport of nurse cell cytoplasm was first suggested by Cooley et al. (1992). However, other nonexclusive mechanisms are possible. It has been proposed, for example, that the force responsible for dumping could reside outside the egg chamber, in the surrounding follicle cells which alter their shape during stage 11, either by squeezing the nurse cells directly, or alternatively by contributing to the elongation of the oocyte which in turn would press against the nurse cells, thus effecting the transfer of cytoplasm into the oocyte (discussed by Spradling, 1993). However, the present results, that a germline deficiency of a myosin II subunit causes a dumpless phenotype, argues strongly that the force for dumping, or at least a major component of that force, is internal to the egg chamber, rather than residing in the somatically derived follicle cells. A failure of cytoplasmic transport might have one of at least three origins: (1) obstruction of the ring canals, (2) absence of the force generating mechanism, (3) lack of a signal to initiate the process. The classic dumpless mutations in the genes chickadee (Cooley et al., 1992), singed (Cant et al., 1994), kelch (Xue and Cooley, 1993), and quail (Mahajan-Miklos and Cooley, 1994), all disrupt structures required to maintain the ring canals free from obstruction. Germline clones of mutations in armadillo (Peifer et al., 1993) so severely perturb the actin cytoskeleton that their dumpless phenotype could conceivably belong to either the first or second category. The dumping problem in sqh1 egg chambers clearly belongs to category no. 2. No mutants have been described which might fall in the third class, candidates for the signalling mechanism, but with the knowledge that myosin II provides the motive force, one can imagine the kind of signalling pathway which might be involved, by analogy with myosin II activation in other nonmuscle systems. Phosphorylation of the RMLC of myosin II on a particular serine residue (serine 19 in vertebrates) is known to trigger smooth muscle contraction and to accompany the rapid contraction of the microfilament network observed during platelet activation (reviewed by Sellers, 1991; Tan et al., 1992). Biochemically, phosphorylation of serine 19 activates both the actin-activated ATPase activity and the ability of myosin to function as a molecular motor (Trybus et al., 1994). In response to a signal, (thrombin binding to its receptor in the case of platelets), a cascade of events is initiated, including a rise in intracellular Ca2+, eventually leading to the activation of the calmodulin-dependent enzyme myosin light chain kinase, which then phosphorylates RMLC at the activating serine, and thus triggers the myosin contraction. (The cytoskeletal changes accompanying platelet activation bear another superficial similarity to nurse cell contraction: the same activating signal initiates a rapid polymerization of actin in a process regulated in part by profilin; see for example Goldschmidt-Clermont et al., 1991). The very high level (80%) of peptide sequence identity of Drosophila RMLC with its vertebrate counteparts (Karess et al., 1991) makes likely the possibility that light chain phosphorylation regulates myosin II activity in this

system as well. A simple model for initating cytoplasmic dumping would have the nurse cells responding to a signal, perhaps a diffusable ligand from the follicle cells, by activating the kinase responsible for RMLC phosphorylation. The kinase phosphorylates the subcortical myosin, thus triggering the contraction of the actin network and resulting in the squeezing of the cytoplasm through the ring canals to the oocyte. The 20% of egg chambers with less than a full complement of nurse cells recalls the cytokinesis defect of zygotic sqh1 larvae, where multinucleate and polyploid mononucleate cells are found at high frequency (Karess et al., 1991). More puzzling are the occasional sqh1 egg chambers containing approximately twice the normal number of nurse cells. In all cases, there is an oocyte at the posterior end, so they cannot be considered tumorous egg chambers, and are unlikely to result from the fusion of two adjacent chambers. Rather, these chambers have apparently sustained a fifth round of cystocyte division. Why this should occur in myosin II-deficient mutants is not clear. However, we have observed in sqh1 zygotic mutants that the third instar larval brains, in addition to their polyploid cells, are also unusually large as a result of abnormal neuroblast proliferation (Karess et al., 1991). That profilindeficient clonal egg chambers display a similar range of nurse cell numbers (Verheyen and Cooley, 1994) suggests that perturbations in actin-myosin cytoskeleton can alter the regulation of mitosis in the early egg chamber. sqh1 is not a null mutation and a small amount of RMLC of wild-type structure is produced in sqh1 clones. Thus the fact that so much of oogenesis proceeds normally in the sqh1 clones does not exclude an essential role for myosin II in earlier stages. Rather it most probably reflects a reduced requirement for myosin II, compared to the massive cytoplasmic movements during dumping. Perdurance of RMLC in wildtype cells from prior to induction of the homozygous sqh1 clones can in principle also contribute to masking the effects of the mutation. However we did not observe any obvious exacerbation of the oocyte phenotype in comparing young (24 days) and old (7-10 days) adult females, or in animals in which the clones had been induced in 1st instar or late 3rd instar larvae (data not shown). Thus the major contribution of residual wild-type myosin II activity is likely to be from the leakiness of the sqh1 allele. Indeed, germline clones of a sqh null allele (P. Jordan and R. Karess, unpublished data), profoundly disrupt oogenesis. The phenotype of the null clones is more complex and agedependent, probably reflecting the effect of diluting out the perduring myosin II. Thus a sqh− clone-bearing ovariole typically has a single stage-14-like egg chamber, and several highly abnormal earlier stages, including apparently ‘dumpless’ chambers, superficially stage 10, but with pycnotic degenerating nurse cell nuclei. The predominant phenotype however is one of greatly reduced cystocyte number (often only one or two large cells per egg chamber) which we interpret as reflecting a failure of cytokinesis in the germarium. The fact that both early and late stage egg chambers are grossly abnormal renders difficult a reliable interpretation of the dumpless phenotype in these sqh− clones. In contrast, the leakiness of the sqh1 allele, by allowing earlier events such as cytokinesis to proceed relatively normally, made it ideal for

Myosin II in oogenesis and early cleavage 1945 observing the specific requirement for myosin II on late events such as dumping. Myosin II and axial expansion The mechanism of axial expansion is unknown. Its dependence on the actin cytoskeleton has been demonstrated by its sensitivity to treatment with cytochalasin D, (Zalokar and Erk, 1976; Hatanaka and Okada, 1991), and by three maternal effect mutations paralog, N441 and N26, which alter the F-actin distribution of unfertilized eggs, and inhibit axial expansion (Hatanaka and Okada, 1991). These studies also revealed the apparent linkage between the mechanism of axial expansion and the posterior cortical contraction of the yolk mass, since the drugs and mutations perturbing one always affect the other. The failure of sqh1 embryos to execute axial expansion of their cleavage nuclei and posterior yolk contraction means that myosin II is critical to these actin-dependent events, and strengthens the mechanistic link between these two processes. The nuclear migration phenotype in sqh1 embryos is strikingly similar to that reported following cytochalasin treatment and with the maternal effect mutations. This observation, combined with the fact that the nuclei in sqh1 embryos are still able to migrate cortically (a movement which depends on microtubules and does not involve the actin cytoskeleton; Baker et al., 1993), are all consistent with a defect specifically linked to a lack myosin II. Our results are also in accord with an earlier observation described in Kiehart et al. (1990) that following injection of anti-MHC antibodies into very early embryos, axial nuclear migration was retarded. One model for axial migration (Foe et al., 1993) holds that the nuclei and their accompanying centrosomes have a natural tendency to expand spherically, but are confined to the narrow central corridor by the cortical actin/myosin network, thus forcing their expansion axially. A second model, based on observations of cytoplasmic movements in living embryos (von Dassow and Schubiger, 1994), argues that cytoplasmic streaming, much like that found in amoebic pseudopods, carries the nuclei along the anterior posterior axis. Recent analysis of Dictyostelium mutants has implicated members of the myosin I class of actin motors in pseudopod movement (Wessels et al., 1991; Titus et al., 1993), and thus presumably in cytoplasmic streaming. Although our results demonstrate an essential role for myosin II in the maintenance of the system responsible for this movement, the participation of other myosin motors is not excluded. Conclusions The product of the gene spaghetti squash, the regulatory light chain (RMLC) of cytoplasmic myosin II, is required within the germline-derived cells of the egg chamber, (the nurse cells and the oocyte) for rapid cytoplasmic transport during oogenesis. Unlike the ‘dumpless’ phenotypes described for other mutations, the actin cytoskeleton of the nurse cells in sqh1 germline clones is not affected, and the failure to dump is not a secondary consequence of plugging the ring canals. Rather, the sqh1 phenotype can be explained as resulting from a reduction in the amount of functional myosin II, which normally generates the force required for the rapid cytoplasmic transport by mediating the contraction of the nurse cells. Similarly, the axial expansion of the syncytial nuclei in the early embryo is blocked in eggs derived from sqh1 maternal

clones, and therefore this process also requires myosin II. Given the important role phosphorylation has been shown to play in regulating myosin II activity in other systems (Sellers, 1991; Tan et al., 1992), it seems plausible to suppose that myosin II phosphorylation will be important in regulating events of oogenesis and embryogenesis as well. We thank Dan Kiehart for the gift of anti-MHC antibody, and Norbert Perrimon for the FRT-FLP stocks. The original observations on axial migration in sqh1 embryos were made by Ms Anca Rosca, then in high school, now in medical school. Thanks to Françoise Poirier, Maryvonne Mevel-Ninio, Pedro Santamaria and Pascale Jordan for their comments on the manuscript. This work was supported by grants to R. K. from the National Science Foundation (USA), the Pew Charitable Trusts (USA), and the Centre National de la Recherche Scientifique (France).

REFERENCES Baker, J., Theurkauf, W. and Schubiger, G. (1993). Dynamic changes in microtubule configuration correlate with nuclear migration in the preblastoderm Drosophila embryo. J. Cell Biol. 122, 113-121. Cant, K., Knowles, B., Mooseker, M. and Cooley, L. (1994). Drosophila singed, a fascin homolog, is required for actin bundle formation during oogenesis and bristle extension. J. Cell Biol. 125, 369-380. Cheney, R. and Mooseker, M. (1992). Unconventional myosins. Curr. Opin. Cell Biol. 4, 27-35. Chou, T.-B. and Perrimon, N. (1992). Use of a yeast site-specific recombinase to produced female germline chimeras in Drosophila. Genetics 131, 643653. Cooley, L., Verheyen. E. and Ayers, K. (1992). chickadee encodes a profilin required for intercellular cytoplasm transport during Drosophila oogenesis. Cell 69, 173-184. Foe, V. and Alberts, B. (1983). Studies of nuclear and cytoplasmic behavior during the five mitotic cycles that precede gastrulation in Drosophila embryogenesis. J. Cell Sci. 61, 31-70. Foe, V., Odell, G. and Edgar. B. (1993). Mitosis and morphogenesis in the Drosophila embryo: point and counterpoint. In The Development of Drosophila melanogaster. (ed. M. Bate and A. Martinez Arias), pp.149-300. Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press. Goldschmidt-Clermont, P., Machesky, L., Doberstein, S. and Pollard, T. (1991). Mechanism of the interaction of human platelet profilin with actin. J. Cell Biol. 113, 1081-1089. Gutzeit, H. (1986). The role of microfilaments in cytoplasmic streaming in Drosophila follicles. J. Cell Sci. 80, 159-169. Hatanaka, K. and Okada, M. (1991). Retarded nuclear migration in Drosophila embryos with aberrant F-actin reorganization caused by maternal mutations and by cytochalasin treatment. Development 111, 909-920. Karess, R., Chang, X.-J., Edwards, K., Kulkarni, S., Aguilera, I. and Kiehart, D. (1991). The regulatory light chain of non-muscle myosin is encoded by spaghetti squash, a gene required for cytokinesis in Drosophila. Cell 65, 1177-1189. Karr, T. and Alberts, B. (1986). Organization of the cytoskeleton in early Drosophila embryos. J. Cell Biol. 98, 156-162. Kiehart, D. and Feghali, R. (1986). Cytoplasmic myosin from Drosophila melanogaster. J. Cell. Biol. 103, 1517-1525. Kiehart, D., Ketchum, A., Young, P., Lutz, D., Alefnito, M., Chang, X.-J., Awobuluyi, M., Pesacreta, T., Inoué, S., Stewart, C. and Chen, T.-L. (1990). Contractile proteins in Drosophila development. Annals New York Acad. Sci. 582, 233-251. Korn, E. and Hammer, J. (1988). Myosins of nonmuscle cells. Annu. Rev. Biophys. Chem. 17, 23-45. Lindsley, D. and Zimm, G. (1992). The genome of Drosophila melanogaster. San Diego, CA: Academic Press. Mahajan-Miklos, S. and Cooley, L. (1994). The villin-like protien encoded by the Drosophila quail gene is required for actin bundle assembly during oogenesis. Cell 78, 291-301. Peifer, M., Orsulic, S., Sweeton, D. and Wiechaus, E. (1993). A role for the Drosophila segment polarity gene armadillo in cell adhesion and cytoskeletal integrity during oogenesis. Development 118, 1191-1207.

1946 S. Wheatley, S. Kulkarni and R. Karess Perrimon, N. and Gans, M. (1983). Clonal analysis of the tissue specificity of recessive female-sterile mutations of Drosophila melanogaster using a dominant female sterile mutation Fs(l)K1237. Genetics 100, 365-373. Sellers, J. (1991). Regulation of cytoplasmic and smooth muscle myosin. Curr. Opin. Cell Biol. 3, 98-104. Spradling, A. (1993). Developmental genetics of oogenesis. In The Development of Drosophila melanogaster. (ed. M. Bate and A. Martinez Arias), pp. 1-70. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press. Tan, J., Ravid, S. and Spudich, J. (1992). Control of nonmuscle myosins by phosphorylation. Ann. Rev. Biochem. 61, 721-759. Titus, M. (1993). Myosins. Curr. Opin. Cell Biol. 5, 77-81. Titus, M., Wessels, D., Spudich, J. and Soll, D. (1993). The unconventional myosin encoded by the myoA gene plays a role in Dictyostelium motility. Mol. Biol. Cell. 4, 233-246. Trybus, K. and Lowey, S. (1988). The regulatory light chain is required for folding of smooth muscle myosin. J. Biol. Chem. 263, 16485-16492. Trybus, K., Waller, G. and Chatman, T. (1994). Coupling of ATPase activity and motility in smooth muscle myosin is mediated by the regulatory light chain. J. Cell Biol. 124, 963-969. Verheyen, E. and Cooley, L. (1994). Profilin mutations disrupt multiple actindependent processes during Drosophila development. Development 120, 717-728. von Dassow, G. and Schubiger, G. (1994). How an actin network might cause fountain streaming and nuclear migration in the syncytial Drosophila embryo. J. Cell Biol. 127, 1637-54.

Warn, R., Gutzeit, H., Smith, L. and Warn, A. (1985). F-actin rings are assoicated with the ring canals of Drosophila egg chambers. Exp. Cell Res. 157, 355-363. Wehland J. and Weber K. (1987). Turnover of the carboxy-terminal tyrosine of a tubulin and means of reaching elevated levels of detyrosination in living cells. J. Cell Sci. 88, 185-203 Wessels, D., Murray, J., Jung, G., J., H. I. and Soll, D. (1991). Myosin IB null mutants of Dictyostelium exhibit abnormalities in motility. Cell Motil. Cytoskel. 20, 301-315. Whitfield, W., Millar, S. Saumweber, H. Frasch, M. and Glover, D. (1988). Cloning of a gene encoding an antigen associated with the centrosome in Drosophila. J. Cell Sci. 89, 467-480: Xue, F. and Cooley, L. (1993). kelch encodes a component of intercellular bridges in Drosophila egg chambers. Cell 72, 681-693. Young, P., Pesacreta, T. and Kiehart, D. (1991). Dynamic changes in the distribution of cytoplasmic myosin during Drosophila embryogenesis. Development 111, 1-14. Young, P., Richman, A., Ketchum, A. and Kiehart, D. (1993). Morphogenesis in Drosophila requires nonmuscle myosin heavy chain function. Genes Dev. 7, 29-41. Zalokar, M. and Erk, I. (1976). Division and migration of nuclei during early embryogenesis of Drosophila melanogaster. J. Microsc. Biol. Cell. 23, 97106. (Accepted 16 March 1995)